Bell's theorem - for or against Hidden Variables?

[gill1109]

Anyway, you're assuming that all four combinations of the a's and b's will occur. That happens in a typical Bell experiment, allowing CHSH inequality to be derived easily. However in Christian's case he instructs us to pick only one (a, b) then use it with each of the experimental data points to produce the script-E terms. 

In his experiment, you observe N particle pairs, get N values of lambda, and then compute any script-E term for any (a, b) you like. For every (a, b): the same N particle pairs, the same N values of lambda.